Crystal and molecular structure of 2-thio-5-carboxymethyluridine and

Wolfgang Hillen, Ernst Egert, Hans Joerg Lindner, and Hans Guenter Gassen .... R. J. Cedergren , David Sankoff , Bernard Larue , Henri Grosjean , Diet...
0 downloads 0 Views
5 3 14

B I 0C H EM I ST R Y

Crystal and Molecular Structure of 2-Thio-5-carboxymethyluridine and Its Methyl Ester: Helix Terminator Nucleosides in the First Position of Some Anticodons? Wolfgang Hillen,* Ernst Egert, Hans Jorg Lindner, and Hans Giinter Gassen

ABSTRACT: The crystal and molecular structures of the odd wobble nucleoside 2-thio-5-methylcarboxymethyluridine (s2mcm5U)and its analogue 2-thio-5-carboxymethyluridine monohydrate (s2cm5U-H20)are reported. Both nucleosides exhibit the )E conformation of the ribose which is connected with low dihedral angles about the glycosidic bonds of 16’ for s2mcm5U and 3’ for s2cmSU, respectively. Although the packing of the molecules is widely different, the location of the bases next to each other is nearly identical, showing a staggered conformation with little base overlap. The geometry of the two carboxymethyl substituents is very similar. Both show the carbonyl oxygen of the carboxylic group located next to C(5) with distances of 2.86 A for s2mcm5Uand 2.85 8, for s2cm5U. M I N D 0 / 3 calculations of the energy minimum of the carboxymethyl substituent reveal that this is the preferred conformation of the molecule and not an artifact resulting from packing forces. Modified uridine derivatives with substituents related to the carboxymethyl group in the wobble position may

thus function as helix terminator nucleosides because steric hindrance prevents the continuation of the anticodon stack toward the 5’ end of the anticodon loop. The molecular structures of these compounds reveal no explanation regarding their restricted wobble recognition. We, therefore, substituted Gm(34) in the anticodon loop structure of tRNAPhewith s2mcm5Uby computer methods in order to study the possible interactions of this nucleoside with its neighbors and the anticodon loop backbone. The location of the substituent at the 5 position in this plot differs from that reported by Berman et al. [Berman, H. M., Marcu, D.. Narayanan, P., Fissekis, J. D.. and Lipnick, R . L. (1978), Nucleic Acids Res. 5 , 893-9031, The molecular aspects leading to this contradiction are discussed in detail. Our plot demonstrates that the slight distortion necessary for the formation of the G.U wobble pair might not be possible for s2mcm5Ubecause of steric hindrance due to the bulky mcm substituent with the backbone of the so-called U turn.

w i t h few exceptions, uridine in the 5’ position of the anticodons of all tRNAs sequenced so far is replaced by modified uridine derivatives. These include thio substitution of the 2-keto oxygen and/or the introduction of various substituents in the 5 position of the uracil moiety (Nishimura, 1972; McCloskey and Nishimura, 1977). Some of these nucleosides amplify the wobble recognition (Crick, 1966). For example, it has been shown that the nucleoside V (uridine-5-oxyacetic acid) in the first position of the Escherichia coli tRNAVa1,and tRNASert anticodons recognizes uracil in addition to adenine and guanine (Takemoto et al., 1973; Ishikura et al., 1971). In contrast, 2-thiouridine derivatives restrict the wobble recognition; they form base pairs with A but not with G (Sekiya et al., 1969). This behavior is presumably due to the reduced ability of sulfur to form a hydrogen bond with HN( 1) of guanine (Yoshida et al., 197 1; Agris et al., 1973). Among the odd widine derivatives present in anticodons are the structures related to 5-carboxymethyluridine (cm5U),I originally isolated from alkaline hydrolysates of bulk yeast and wheat embryo tRNAs (Gray and Lane, 1967, 1968). Further studies revealed that cmsU derivatives also occur in different tRNA species. The methyl ester

(mcmjU) has been found in yeast t R N A (Tumaitis and Lane, 1970) and the amide in yeast and in wheat embryo tRNAs (Dunn and Trigg, 1975). The crystal structure of the latter has been recently reported (Berman et al., 1978). Its 2’Gmethyl derivative has been isolated from yeast t R N A hydrolysates (Gray, 1976). It has been shown that mcm5U occurs in the wobble position of tRNAArg, from brewer’s yeast (Kuntzel et al., 1975). The analogue of mcm5U, in which the 2-keto oxygen is replaced by sulfur (s2mcm5U),has also been isolated from yeast tRNA hydrolysates (Baczynskyi et al., 1968; Kwong and Lane, 1970) and was found in the wobble positions of tRNALYS2(Madison et a]., 1972) and tRNAGIu3(Kobayashi et al., 1974) from baker’s yeast. Sekiya et al. (1969) have shown by the ribosomal binding assay and by incorporation of Glu in in vitro synthesized hemoglobin, that tRNAGIU3recognizes only GAA as a codon and not GAG. Recent studies of the coding properties of tRNAArg3 from brewer’s yeast also revealed that the sulfur lacking mcm5U recognizes only AGA but not AGG (Weissenbach and Dirheimer, 1978). Therefore, it seems that in these wobble nucleosides the methyl ester of the carboxymethyl group in the 5 position of uridine rather than the 2-thio group causes the altered recognition pattern of the respective tRNAs. This behavior of mcm5U was predicted by Sen and Gosh (1976) from desulfination of tRNALbS2from yeast. Studies of the biosynthesis of mcm5U support the conclusion that cm5U is the precursor of its blocked derivatives because cm5U residues in bulk t R N A can be methylated by a cell-free extract of wheat embryo (Bronskill et al., 1972). It is not clear as yet whether the methyl ester and the amide are alternative pathways or if they are metabolically interconvertible (Dunn

From the Fachgebiet Biochemie and Fachgebiet Organische Chemie I (E.E. and H.J.L.). lnstitut fur Organische Chemie und Biochemie. Technische Hochschule Darmstadt, Petersenstr. 2 2 , 6100 Darmstadt, Federal Republic of Germany. Receiced April 27, 1978. This work was supported by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. Abbreviations used: s2cm5C. 2-thio-5-carboxymethyluridine: cm5U, 5-carboxymethyluridine; s2mcm5U, 2-thio5-methylcarboxymethyluridine;mcm5U, 5-methylcarboxymethyluridine; ncm5U, 5-carbamoylmethyluridine; t6A, N6-(N-threonylcarbonyl)adenosine; g6A, N6-(N-glycylcarbonyl)adenosine. +



0006-2960/78/0417-5314$01.00/0

0 I978 American Chemical Society

2-TH IO- 5 - C A R

B O X Y M E T H Y L U R ID1 N E M E T H Y 1.

ESTER

VOL.

17,

NO.

24, 1978

5315

and polarization factors but not for absorption. 71 3 reflections of s2cm5U.H20 had IF1 > 3 u ~ and , 460 reflections of s2mcm5U had IF1 > 2 u ~ . s2cm5U.H20 s2mcmW The crystal structures were solved with SHELX-76 (Shelmol form. C Ii H 1 4 0 7 k S * H 2 0 C I ~ H I ~ ~ ~ N Z S drick, unpublished program). Isotropic refinement reduced mol wt 336 332 R to 0.1 13 (s2cm5U.H20) and 0.1 19 (s2mcm5U), which 0.2 X 0. I X 0.8 0.1 X 0.02 X 0.6 cryst dimens (mm) dropped to 0.084 for s2cm5U-H20with anisotropic temperaspace group p212121 p212121 a = 19.27 ( I ) a = 23.15 ( I ) cell const (A) ture factors, whereas only the sulfur atom was made anisob = 11.571 ( 6 ) b = 14.006 (8) tropic for s2mcm5U because of the low number of measured c = 5.320 ( 7 ) c = 5.316 ( 5 ) reflections. Successive electron-density difference maps yielded (A3) 1424.2 1435.7 the positions of 10 of the 16 hydrogen atoms in the respective 4 4 Z structures. The six remaining hydrogen atoms were almost Pobsd ( g c m - 3 ) I .57 1.53 exclusively bonded to carbon atoms and could be positioned Pcalcd (g’cm-3) 1.567 1.536 23. I 22.4 u geometrically. The parameters of the hydrogen atoms were not varied. The refinement converged a t R values of 0.078 and 0.097 for s2cm5U.H20 and s2mcm5U,respectively. Parameters and structure factors are available as supplementary materiand Trigg, 1975). However, the carboxymethyl derivative itself al. has not been found in tRNAs. In order to explain the unique coding properties of mcm5UThe calculation of the preferred conformation of the carboxymethyl group was carried out with a revised version and s2mcm5U-containing tRNAs and the absence of cm5U in (Bischof, 1976) of the S C F - M O method M I N D 0 / 3 , which tRNA, we determined the crystal structures of both s2mcm5U and s2cmSU. considers all valence electrons and minimizes the total energy with respect to the geometrical variables (Bingham et al., C 1975).

T A B L E I : Crystal D a t a .

i c

o w

Results Crystal data of the two nucleosides summarized in Table I show that both crystallize in the same orthorhombic space group P212121 with needle axes c. Conformation of s2cm5Uands2mcm5U.The bond lengths and angles (Figure 1) agree with those of similar nucleosides, but a detailed discussion is impossible due to the increased standard deviations which are the result of the very small crystals obtained from these compounds. The pyrimidine rings are, within experimental error, planar with a standard deviation of the ring atoms of u = 0.01 A. Only the sulfur atom shows a perceptible deviation (0.1 1 A) from the base plane in s2mcm5U,whereas the bond to C( 1’) is much less bent than in s2cm5U where C( 1’) deviates by 0.22 A. The most striking molecular feature is the remarkable agreement between the conformations of the substituents in the 5 position (Figure 2). I n both molecules the bond C(51)C(52) is approximately perpendicular to the base plane and the carbonyl oxygen of the carboxylic group is situated next to the ring atom C(5), whereas the carboxyl oxygen points away from the base showing dihedral angles C(5)-C(51)C(52)-O(52) of 163’ and 166’ for s2cm5U and s2mcm5U, respectively (Figure 1 ). The ribosyl moieties have the C(3’)-endo conformation with C( l’), C(2’), C(4’), and O( 1’) in the plane (u = 0.03 and 0.01 A), while C(3’) deviates by 0.55 and 0.44 A for s2cm5Uand s2mcm5U, respectively. As usual in nucleosides, the C ( 1’)O( 1’) bonds are shorter than C(4’)-O( 1’). The orientation a t the glycosidic bond N ( 1)-C( j’) is anti with dihedral angles C(6)-N(l)-C(l’)-O(l’) of x = 3’ (s2cm5U) and 16’ (s2mcm5U), which are in the range between 0’ and 20’ typical for 2-thiopyrimidines (Lin and Sundaralingam, 197 1; Lin et al., 1971; Kojic-ProdiC et al., 1974, 1976; Hawkinson, 1977) and fit in the common relation between the puckering of the ribose and the orientation of the base (Sundaralingam, 1969; Egert et al., 1977). The two nucleosides differ, however, in the conformation around C(4’)-C(5’): s2mcm5Ushows the usually preferred gauche-gauche arrangement with dihedral angles 0( 1’) - C (4’) - C (5’) -0(5’) and C( 3’) -C(4’) -C (5’) -0(5’) of -7 1 O and 47O, respectively, which allows the intramolecular interaction C(6)H.s .0(5’) (Table 11) often found in nucle-

How A0

OH

I n addition, a calculation of the conformational properties of the carboxymethyl substituent in uridine using the M I N D 0 / 3 method to optimize the geometry was carried out to give further support to the assumption that the crystalline and solute conformations are identical. Furthermore, we computed a plot of the U turn from the crystal structure of tRNAPhe(Quigley and Rich, 1976) in which we replaced the 2’-O-methylguanosine(34) by s2mcm5U to demonstrate the possible interactions of substituents a t the 5 position in modified uridine nucleosides with their neighboring nucleotides and the backbone of the anticodon loop, assuming that the so-called U turn is a general structural feature of all anticodon loops. Materials and Methods 2-Thio-5-carboxymethyluridine methyl ester (s2mcm5U) was a generous gift of Dr. Vorbriiggen, Schering AG., Berlin. After several unsuccessful attempts with different solvents, one very small needle-shaped crystal was obtained from a mixture of acetone: methanol, and chloroform ( I :I :2, v/v), which was allowed to evaporate slowly from an open vessel a t ambient temperature. As soon as the first tiny crystals formed a t the interface of the solvents, the vessel was tightly stoppered and slowly cooled to 10 ‘C. Treatment of s2mcm5U with formic acid/water ( l : l , v/v) yielded clusters of thin needles of 2thio-5-carboxymethyluridine monohydrate (s2cm5U-H20). The intensities were collected on a Stoe two-circle diffractometer (Cu K a radiation) equipped with a graphite monochromator. Both crystals were oriented along c . In this way, 741 symmetry-independent reflections of s2cmSU.H20 and 555 of s2mcm5Uwith 0 160’ were measured in the 0 - 20 scan mode. The data were corrected for background and for Lorentz

5316

BIOCHEMISTRY

HILLEN

a

ET A t

P

a

b

F I G U R E 2:

The structural representation of (a) s2cm5Uand (b) s2mcm5C:

with the thermal ellipsoids of the heav) atoms (only i n l a ) .

of Intermolecular Hydrogen Bonds of s’mcm5U. Including the Characteristic Intramolecular C(6)-H.. - 0 ( 5 ’ ) Interaction.

T A B L E I I : List

9: X - H - Y 0:3’)

I : The bond lengths and the bond and dihedral angles of (a) s?cm’U (ax, = 0.02 A,o x x x= l o s x = C, 0. N , S) and (b) s2mcm5U( u x x = 0.03 K . oXIx = 2 ” ) . I-IGURE

osides (Egert et al., 1978), whereas s2cm5U exhibits a gauche-trans conformation with dihedral angles of 58’ and 176O, which was already found for some other 2-thiouridine derivatives (Kojic-Prodie et al., 1974, 1976; Hawkinson, 1977). The gauche-trans conformation seems to be mainly induced by packing forces. Packing of the Molecules. A view down the c axis (Figure 3) illustrates the crystal structures which are determined by three screw axes parallel to the edges of the cell. The packing of s2mcm5ti shows no definite separation of base and ribose regions. This arrangement is stabilized by one intermolecular hydrogen bond N(3)-H. .0(5’), whereas interactions between riboses do not occur (Table 11). The three-dimensional structure of s2cm5U.H20can be divided into layers parallel to the bc plane showing the well-known sequence: ribose-base.

-

X-Ha .Y e

N(3)-H** * O w ) C(6)-H** * 0 ( 5 1 )

X-H

(A)

I. I I. I

H.. .Y

I .7

(A) X . . .Y (A)

(deg)

I55

0(5’)-H...0(5l) I ?

2.4 2. I

2.78 3.38 2.76

143 112

C(6)-H** .0(5’)

2.4

3.41

I46

I. I

base-ribose. -ribose-base. .base-ribose, etc. Next to an interbase hydrogen bond N(3)-H.. 6 ( 2 ) adjacent molecules are connected predominantly by a water molecule which participates in six hydrogen bonds (Table 111). The water oxygen exhibits a tetrahedral alignment of two strong accepting and two donating hydrogen bonds, one of which is trifurcated. All together, the water molecule is linked with four bases and two ribosyl moieties and thus plays the dominant role in the intermolecular arrangement of the molecules (Figure 4). I n spite of the rather different crystal structures, the base stacking patterns of s2cm5U and s2mcm5U are nearly identical (Figure 5 ) . The bases are packed along the c axis at a distance of 3.60 (s2cm5U) and 3.44 A (s2mcm5U), showing a staggered

2 - T H I O - 5 - C A R B O X Y M E T H Y L U R l D l N EM E T H Y L E S T E R

VOL. 17, NO. 24, 1 9 7 8

5317

F I G U R E 4: The dominant influence of the water molecule (hatched atoms) on the crystal structure of s2cm5U.H20. The hydrogen bonds are indicated by dotted lines.

F I G U R E 3: View down the c axes of (a) s2cm5U.H20 and (b) s2mcm5U with a horizontal and b vertical. The individual base stacks are designated with letters. They are in relation with the following symmetry operations: (A B) screw axis parallel to c; ( A C) screw axis parallel to a ; ( A c1 D) screw axis parallel to b. The hydrogen bonds are indicated by dotted lines.

-

-

T A B L E I l l : List of Intermolecular Hydrogen Bonds of s2cm5L.

H20. 4 X-H-Y

X-H.**Y

X-H

(A) H * . * Y(A) X*..Y(A)

(deg)

1.0

2.4 2.4 2.3 2.2 I .8

3.27 3.49 3.52 3.20 2.82

I58 157 169 163 I80

1.3 1.3 1.3 1.0 1.3 1. I

3.0 2.3 2.2 1.9 I .4 I .7

3.66 2.96 3.3 I 2.90 2.70 2.76

112 1 IO 146 I80 I64 I80

N(3)-H. 4 ( 2 ) C(6)-H***0(51) C(3’)-H* * mO(5’) 0(3’)-H.*.0(5’) O(S’)-H. * *0(51)

0.9 1.2 1.2

O(W)-H**.S(2) O ( W ) - H . * .0(4) O(W)-H*..0(52) O(W)-H’*..0(3’) 0(52)-H***O(W) 0(2’)-H. * *O(W)

1.1

F I G U R E 5: Views perpendicular to thc base planes of (a) s2cm5U.H20 and (b) s2mcm5Ushowing the staggered arrangement of the six-membered rings and the intermolecular interactions (dotted lines). The atoms of the middle one out of thc three nucleosides are hatched.

arrangement. This arrangement is stabilized by the hydrogen bond 0(5’)-H. .-0(51) and the weak interaction C(6)-H. aO(51) in both nucleosides. s2cm5Uexhibits a further stabilization by the water molecule which connects the carboxyl groups of adjacent bases by means of two hydrogen bonds. Calculation of the Carboxymethyl Group Conformation. The nearly identical conformation of the substituents a t the 5 position of s2cm5U and s2mcm5U gave rise to theoretical

-

calculations using the MIND0/3 method which should clarify whether this conformation represents an energy minimum because of electronic and steric reasons or if it is mainly induced by packing forces. We chose c m W as our model compound and not the 2-thio derivative because it fits better into the series of uridines substituted a t the 5 position, the conformational properties of which we have calculated. The ribose was replaced by the 1hydroxyethyl group, -CH(OH)CH3, which has a similar electronic influence on the base, which was verified by calculations of complete nucleosides and series of simplified models (manuscript in preparation). The optimization included all bond lengths except the bonds to hydrogen atoms (which were kept a t 1.08 A) and all bond angles except those of the 1hydroxyethyl group (which were given the ideal tetrahedral angle of 1 0 9 . 5 O ) . Furthermore, a free rotation of the hy-

531 8

B I O CH E M I S T R Y

H I 1 1 I-N

m a % 2 5 C I - I - L L - I - -

+

t

I

O M V

f

F I G U R E 6 : A plot of the energy relative to the global minimum at 340' vs. CY [the dihedral angle C(5)-C(Sl)-C(52)-0(5l)] with sawhorse

projections of the conformations belonging to the extrema. The calculated values marked by asterisks are fitted by a polynomial approximation. The experimental values of a are 339' for s2cm5L and 35 1 for s2mcmsU.

droxyethyl group was permitted, leading to a dihedral angle C(6)-N( 1)-C( 1')-O( 1') of about 50". The carboxymethyl group has three degrees of rotational freedom around the bonds C(S)-C(51), C(51)-C(52), and C(52)-O(52). The conformation of the latter, which defines the orientation of the carboxyl hydrogen H(52) with respect to 0(51), stayed at the starting value of 0" for the dihedral angle 0(51)-C(52)-0(52)-H(52). A free rotation around C(5)-C(5 1) is strongly hindered by repulsing interactions between the carboxymethyl group and the ring substituents O(4) and H(6), resulting in a high activation energy. Therefore, only two small ranges of this angle represent the favored conformations, which require the C(5 I)-C(52) bond approximately perpendicular to the six-membered ring with C(52) "above" or "below" the base plane. Following these arguments, the dihedral angle C(4)-C(S)-C(5 I)-C(52) was fixed a t -90". The conformational freedom of the carboxymethyl group is now reduced to the rotation of the carboxyl group around C(51)-C(52). Therefore, we performed calculations varying the dihedral angle cy C(5)-C(51)-C(52)-0(51) from 0" to 360" in 30" steps and optimized all 34 variables (bond lengths, bond angles, and dihedral angles) mentioned above. The results are shown in Figure 6. The plot of the energy vs. a exhibits two minima a t 180' and 340" separated by two maxima at 80" and 250". The minima differ by only 0.1 kcal/mol and correspond to conformations with one oxygen of the carboxyl group "above" C(5), as indicated in Figure 6. The energy of the maxima is 2.1 and 1.8 kcal/mol higher than the energy of the global minimum at 340'. Discussion The crystal structures of s2cmsU.Hz0 and s2mcmsU reveal two remarkable similarities: the confoimation of the molecules and the base stacking. Although the packing of these compounds is different (Figure 3), the location of the base moieties next to each other is nearly identical (Figure 5 ) , and no noticeable base overlap occurs in the crystals. In spite of its considerable polarizability, the 2-thio group is not a t all involved in the base-base interaction as in 2-thiocytidine (Lin et al., 1971). However, it is not possible to deduce the base overlap of nucleosides within nucleic acids from their crystal structures because the energy minimum of the packing in the crystal is probably found by optimizing other interactions, for example, intermolecular hydrogen bonds. Both nucleosides exhibit the C(3')-endo conformation of the

FT 41

ribose combined with small dihedral angles over the glycosidic bonds. If this conformation is also the favored one in aqueous solution, stacking interactions in these nucleosides would be more stable because in stacked ribonucleic acids the C(3')-endo conformation is exclusively found (for a recent review of this subject, see Kallenbach and Berman, 1977). The results from the M I N D O / 3 calculations (Figure 6 ) show the preferred orientation of one oxygen of the carboxyl group either "above" or "below" C(5). The distances between O(51) and C(5) are only 2.85 and 2.86 A for s2cmsU and s'mcm'U, respectively. The energy curve of the rotation around C(51)-C(52) in Figure 6 can be explained by a simple MO model which considers the interaction between the orbitals of the carboxymethyl group and those of the C(4)=0(4) double bond. Detailed studies reveal two possible interactions: ( 1) a bonding (Le., stabilization) interaction between the lone pairs of one of the oxygens from the carboxyl group and the T * orbital which has its greatest coefficient at C(4) end (2) an antibonding (ix., destabilization) interaction between the same lone pairs of this oxygen and the K orbital localized mainly at O(4). These opposite effects determine the preferred conformation of the carboxymethyl group. The orbitals of the carbonyl oxygen. 0 ( 5 I ) %in the carboxylic group have the higher energ) and are better adjusted for an overlap with the carbonyl group C(4)=0(4). This leads to the more pronounced stabilizing (a = 340") and destabilizing ( a = 80') interactions when this oxygen is next to C(5) as compared to O(52) ( a = 180' and 250°, Figure 6). Therefore, the carboxymethyl substituent has its energy minimum a t the balanced distance of 0 ( 5 I ) as close to C(4) and as far away from O(4) as possible. The experimental values for a obtained from the crystal structures are 339' for s2cm5U and 35 1" for s2mcm5U and agree with the calculations as well as the values for (Y of 16. I ' for cm5U and 329.2" for ncm5U reported by Berman et al. (1978). The empirically found conformation. also derived only from electronic and steric properties by the M I N D 0 / 3 approximation, is also valid for the carboxyl-blocked derivatives like the methyl ester or the amide and for the respective 2-thio compounds. Thus, these results support the assumption that the molecular conformation found in these crystals is a common feature of odd uridines containing substituents at the 5 position related to the carboxymethyl group. The molecular structure of s2mcm51J does not explain the influence of this substituent at the 5 position on the reduced wobble recognition of either s2mcm5Uor mcm51J in tRNAs (Sekiya et al.. 1969; Weissenbach and Dirheimer, 1978). Therefore, this finding can most likely be explained by the sterical arrangement of these odd nucleosides within the anticodon loop. Assuming that the crystal structure of tRNAPhe represents the biologically active conformation, we substituted Gm(34) with s2mcmiU by computer methods using the Cartesian coordinates of the anticodon loop (Quigley at al., 1975). A similar procedure was used by Parthasarathy et al. (1977) to discuss the influence of thA and ghA on the anticodon loop structure. Recently, Gm(34) was substituted with the molecular structure of 5-carbamoylmethyluridine (Berman et al., 1978). These authors postulated a hydrogen bond from the carbonyl oxygen 0 ( 5 I ) to the 2'-hydroxyl group of the ribose from U(33) fixing the location of this wobble nucleoside. The analog plot for s2mcm5Uwith the bond lengths, bond angles, and dihedral angles taken from the crystal structure is shown in Figure 7. The interaction of O(51) with OH(2') from U(33) is indicated by a line. The distance between the oxygens in this structure would be 2.05 A, which is much less than the sum of the van der Waals radii. If one rotates the base around the glycosidic bond, lowering X(:N to S0--the value found in the

2.THIO-5-CARBOXYMETHYLURlDlNE M E T H Y L ESTER

VOL.

17,

NO.

24, 1978

5319

U

A stereo picture of the anticodon loop of tRNAPheshowing U(33) and the three anticodon nucleosides with Gin(34) replaced by s2mcm5U. The hydrogen bond from N(3) of U(33) to phosphate(36) stabilizing the U turn and the interaction of the carbonyl oxygen O(5 I ) with the 2’-OH group from U ( 3 3 ) are indicated by thin lines. FIGURE7 :

F I G U R E 8: A stereo picture with the same atoms and the same orientation as in Figure 7 but with the substituent at the 5 position of U(34) rotated by 180’. resulting in an interaction of the mcrn group with the phosphodiester(34) between U(33) and s2mcmsU (34), which is indicated by thin lines.

crystal structure of ncm5U-the distance is 2.25 A, which is still less than usually found for hydrogen bonds. Even if one decreases XCN to O’, the distance (2.41 A) is still shorter than a hydrogen bond. In addition, the angle 0(2’)-H. sO(51) would be approximately loo’, which does not favor hydrogen bonding. These values indicate that the interaction of the mcm substituent with the ribose of U(33) must be repulsive. In order to find a more stable arrangement of this nucleoside in the anticodon loop, we rotated the mcm substituent by 180’ around C(5)-C(5 1 ) to reach an equivalent energy minimum on the other side of the six-membered uracil ring next to C(5). The resultant conformation of the carboxylic group was found in the crystal structure of cm5U (Berman et al., 1978). The plot of the anticodon loop derived in this manner is shown in Figure 8. The interaction of the mcm group with the phosphodiester between U(33) and Gm(34) is indicated by lines between the carboxylic oxygens and one phosphate oxygen. These distances, derived from the replacement of Gm(34) by s2mcm5U without any distortion, are 2.15 and 2.97 A, which are equal or only a little less than the sum of the van der Waals radii of these atoms. Therefore, this nucleoside might cause a slight deviation from this arrangement, which would bring the 2-thio group next to N( 1) of a neighboring uridine, which is in position 35 of the tRNACIu3and tRNALySZsequences from baker’s yeast, allowing the preferred S(2)-N( 1) stacking interaction (Mazumdar et al., 1974). The U-G base pair is formed via hydrogen bonds between the 2-keto and 3-imino groups of the uridine moiety and the 1-imino and 6-keto groups of guanosine. This formation requires a slight rotation of the uridine residue within the anticodon stack toward the backbone of the U turn. Assuming that

-

the anticodon loop conformation found in the crystal undergoes a codon-anticodon interaction, this rotation would be impossible for mcm5U in the wobble position because of the steric hindrance of the mcm group, as discussed above. This bulky substituent in the 5 position may, for geometric reasons, prevent the G-U wobble pair. Also, the neighborhood of the negatively charged phosphodiester group could explain why cm5U is not found in tRNAs. The small distance between these two negatively charged groups would result in an enhanced Coulomb repulsion and thereby disturb the helical structure of the anticodon stack. Odd uridine derivatives containing one of these substituents a t the 5 position in this conformation may be called “helix terminator nucleosides” because they always terminate a right-handed helix of stacked nucleic acids. The adjacent nucleotide in the polynucleotide chain cannot stack on these uridine derivatives due to the steric hindrance of the substituent a t the 5 position. This behavior also agrees very well with their occurrence a t the 5’ end of the anticodons in some tRNAs next to the so-called U turn (Quigley and Rich, 1976) (Figures 7 and 8), where they may contribute to this possible common structural feature of the anticodon loops by preventing the stack of U(33) on the anticodon helix (Figure 8). Such a mechanism might be especially important for 2-thiouridine derivatives because 2-thiouridine is known to exhibit a highly thermostable stacking interaction in the homopolynucleotide (Bahr et al., 1973). Acknowledgements We thank Dr. P. Bischof for providing us with his computer program and M . C. Boehm for useful discussions. Further-

5320

BI

ocH E M I STRY

more, we express our gratitude to E. Ronnfeldt and 0.E. Beck for their help in preparing the manuscript. Supplementary Material Available Thermal parameters, the observed and calculated structure factors, and the positional parameters with their estimated standard deviations ( 1 1 pages). Ordering information is given on any current masthead page. References Agris, P. F., Soli, D., and Seno, T. (1973), Biochemistry 12, 4331-4337. Baczynskyi, K., Biemann, K., and Hall, R. H. (1968), Science 159, 1481-1483. Bahr, W., Faerber, P., and Scheit, K.-H. (1973), Eur. J . Biochem. 33, 535-544. Berman, H. M., Marcu, D., Narayanan, P., Fissekis, J. D., and Lipnick, R. L. (1978), Nucleic Acids Res. 5, 893-903. Bingham, R. C., Dewar, M. J. S., and Lo, D. H. ( 1 9 7 9 , J . Am. Chem. SOC.97, 1285- 1293. Bischof, P. (1976), J . Am. Chem. SOC.98, 6844-6849. Bronskill, P., Kennedy, T. D., and Lane, B. G. (1972), Biochim. Biophys. Acta 262, 275-282. Crick, F. H. C. (1966), J . Mol. Biol. 19, 548-555. Dunn, D. B., and Trigg, M. D. M. (1975), Biochem. Soc. Trans. 3, 656-659. Egert, E., Lindner, H . J., Hillen, W., and Gassen, H. G. (1977), Acta Crystallogr., Sect. B 33, 3704-3707. Egert, E., Lindner, H . J., Hillen, W., and Gassen, H. G . (1978), Acta Crystallogr., Sect. B 34, 2204-2208. Gray, M. W. (1 976), Biochemistry 15, 3046-305 1. Gray, M. W., and Lane, B. G. (1967), Biochim. Biophys. Acta 134, 243-257. Gray, M. W., and Lane, B. G . (1968), Biochemistry 7, 3441 -3453. Hawkinson, S . W . (1977), Acta Crystallogr. Sect. B 33, 80-85. Ishikura, H., Yamada, Y., and Nishimura, S. (1971), Biochim. Biophys. Acta 228, 47 1-48 1. Kallenbach, N . R., and Berman, H. M. ( 1 977), Q. Reu. Biophys. I O , 138-236.

1iILL.F.N E T A L

Kobayashi, T., Irie, T., Yoshida, PA., Takeishi, K.. and IJkita, T. ( 1 974), Biochim. Biophys. izc-tn 366, 168 18 I . KojiC-ProdiC, B.. Liminga, R., SljukiC, M., and Ruiii.-ToroS, Z. (1974), Acta Crystallogr. Sect. B 30, 1550- 1555. KojiC-ProdiC, B.. Kvick, A,, and RuiiC-ToroS, 2,.( 1 976), Artn Crystallogr. Sect. B32. 1090- 1095. Kuntzel, B., Weissenbach, J., Wolff. R. E.. 'Tumaitis-Kennedy. T. D., Lane, B. G., and Dirheimer, C. ( I 975). Biorhimie 57, 61 --70. Kwong. T . C., and Lane. B. G. ( I 970). Biochim. Biophys. Acta 224, 405-4 12. Lin, G . H.-Y., and Sundaralingam, M. (1971). Acta Crystallogr. Sect. B 27, 961 -969. Lin, G . H.-Y., Sundaralingam, M . , and Arora, S. K. ( I 971), J . A m . Chem. Soc. 93, 1235.- 1241. Madison, J . T., Boguslawski, S . .I., and Teetor, G . €-I. (1972), Science 176, 687-689. Mazumdar, S. K., Saenger, W., and Scheit, K. H. (1974), J . Mol. Biol. 85, 213-229. McCloskey, J. A,, and Nishimura, S. ( 1977), Ace. Chem. Res. I O , 403-410. Nishimura, S. ( 1 972), Prog. Nucleic Arids Res. Mol. Biol. 12, 50--85. Parthasarathy, R., Ohrt, .I.M., and Chheda, G . B. (1977), Biochemistry 16, 4999-5008. Quigley, G. J., and Rich, A. (1976), Science 194, 796-806. Quigley, G. J., Seeman, N. C., Wang, A. H., Suddath, F. Id., and Rich, A. (1975), Nucleic Acids Res. 2, 2329-2341. Sekiya, T., Takeishi, K., and Ukita, T. (1969). Biochim. Bioph,vs. Acta 182, 41 1 --426. Sen, G . C., and Gosh, H. P. (1976), Nucleic Acids Res. 3, 523-535. Sundaralingam, M. (1969), Biopolymers 7 , 821-860. Takemoto, T., Takeishi, K., Nishimura, S., and Ukita, T. ( 1 973), Eur. J . Biochem. 38, 489 -496. Tumaitis, T. D:, and Lane, B. G. (1970), Biochim. Biophys. Acta 224, 391-403. Weissenbach, J., and Dirheimer, G. (1978), Biochim. Biophys. Acta 518, 530-534. Yoshida, M., Takeishi, K., and Ukita, 'T.(1971). Biochim. Bi0phy.Y. Acta 228, 153- 166.